NVIDIA GTX 595 (picture+Details)

[XRL8]

You gotta miss the "Detonators" moniker of nv drivers a while back in a situation like this.

[saaya]

All I see is one card unusable @stock (HD 6990, even louder than 2 GTX 480 in SLi) and another one (GTX 590, cool noise and approx. same temps than HD6990) that is usable @stock.
With proper drivers and soft voltage modding, the card oc quite well.
But, you're certainly right, power regulation seems insufficient for massive OC.

Anyway, Damien take another thermal shoot, more precise :

the mainboard is 80-90C hot?
and the scale goes to 90C and then the vga is 110C? 0_o
i think the didnt calibrate the thing properly tbh...

[Olivon]

the mainboard is 80-90C hot?
and the scale goes to 90C and then the vga is 110C? 0_o
i think the didnt calibrate the thing properly tbh...

Damien uses a Sonata III for his protocol so temps are really hot inside

[Mungri]

All of this multi-GPU/single slot stuff is nonsense. I would never buy a 6990 nor a 590. If you want multi-gpu, just get multiple single cards and a proper motherboard to support them. It works out better in every way.

Yup. Each card runs cooler and quieter than a dual GPU card can, and with higher clock speeds.

[Manicdan]

maybe they should be using triple slot cooling for such behemoths,

honestly, its one less slot than 2 gpus, and might actually keep it cooler if they have so much extra room for fins and the fan would have plenty of breathing room.

hearing a 700$ card die is just plain stupid. way too many attempts to prevent cards from being used in things like furmark/occt, and not enough in providing stable power delivery.

[saaya]

Damien uses a Sonata III for his protocol so temps are really hot inside

it doesnt matter what case he uses, the temp probe isnt calibrated well...

[DarthShader]

Anyway, Damien take another thermal shoot, more precise :

So the 111C is in the middle of the card... exactly where the VRMs are. Explains a lot.

nVidia screw with their OCP software solution, but with proper driver version things seems okay.

Nope, cards die with new drivers too. See reviews thread for sme references.

it doesnt matter what case he uses, the temp probe isnt calibrated well...

Write him an e-mail and teach him how to do it properly.

[h0bbes]

it doesnt matter what case he uses, the temp probe isnt calibrated well...

You can't tell if it is calibrated or not just by looking at that scale.

[saaya]

You can't tell if it is calibrated or not just by looking at that scale.

true... you need common sense to tell that theres something wrong

[Manicdan]

or he tries to use a common scale for all gpus so you can look at past photos and compare without having to first calibrate your eyes to each image

[sam3]

yeah but... that still doesnt explain why the card blew...
he increased voltage from 1v to 1.2v and used stock clocks... and that blew the card? then the pwm seems to be pretty whimpy, no?

if it blows almost instantly with 1.2v and stock clocks, how safe is it to overclock the card and run it at 1.1v? how long until it pops?

...................

The youtube video says the voltage was at 1.025 V when it blew

GPU Clock @ 772 Mhz
GPU VCore @ 1.025 V

http://www.youtube.com/watch?v=sRo-1VFMcbc

Edit: Ok now I realised you were talking about the techpowerup article and not the quoted SweClockers.com link

[SKYMTL]

The youtube video says the voltage was at 1.025 V when it blew

GPU Clock @ 772 Mhz
GPU VCore @ 1.025 V

http://www.youtube.com/watch?v=sRo-1VFMcbc

I just wanted to remind you that several people have reported that setting 1.025V in software actually pushes about 1.1V to the GPUs as measured by a DMM.

[scottsche]

I swear next time nvidia should just send out 50 cards to random pc enthusiasts via a lottery, instead of sending them to reviewers, and tell them to post their impressions of the STOCK card on a forum. I WISH i could drop $700 for one of these cards. If I got one for free I wouldn't be finding every flaw with it, I would be jumping up and down saying THANKS. It would be a different matter entirely if I was gaming and the card blew. life happens. That's what RMA and a lifetime/3 year warranty is for, it is not there so you can push a card to the breaking point and whine that nvidia makes a sucky card, and then take advantage of the system to get yourself a new one.

Hey nvidia if you are listening, send me a GTX590 so I can GPU render with it and post my impressions of what this card can do in capable hands.

Drives me nuts thinking about how the best, most power efficient GF 110's went into these 590's and people are wasting them by OCing when they don't know what they are doing. The 590 is clearly to any reasonable person a card that is already pushing the limit of what can be cost effectively engineered at this point in time, so why expect 30% OC on a card that is already pushed to the limit when you pull it out of the box.

And someone send some case fans to France, ffs, they seem to have run out.

[InCredible]

the mainboard is 80-90C hot?
and the scale goes to 90C and then the vga is 110C? 0_o
i think the didnt calibrate the thing properly tbh...

im no expert but the color range goes higher then the degree range. You see the little line on the color range..thats 90..it goes higher then that..therefore the temp COULD be correct. Despite that cards are still dying so that leads me more to believe the temp probe is correct and would correlate to why this may be happening. Thats just my common sense at work..

[Postmodum]

I've been reading that water cooled cards ( i'm talking about classified hydro copper with back plate ) are handling 840Mhz with 1.05v just fine ... and that is a really good overclock!
It seems that it's on the 50's at full ... amazing stuff ... if the card doesn't throttle down due to the drivers ... it must be a monster and enter the 580 SLi realm ...

[InCredible]

I swear next time nvidia should just send out 50 cards to random pc enthusiasts via a lottery, instead of sending them to reviewers, and tell them to post their impressions of the STOCK card on a forum. I WISH i could drop $700 for one of these cards. If I got one for free I wouldn't be finding every flaw with it, I would be jumping up and down saying THANKS. It would be a different matter entirely if I was gaming and the card blew. life happens. That's what RMA and a lifetime/3 year warranty is for, it is not there so you can push a card to the breaking point and whine that nvidia makes a sucky card, and then take advantage of the system to get yourself a new one.

Hey nvidia if you are listening, send me a GTX590 so I can GPU render with it and post my impressions of what this card can do in capable hands.

Drives me nuts thinking about how the best, most power efficient GF 110's went into these 590's and people are wasting them by OCing when they don't know what they are doing. The 590 is clearly to any reasonable person a card that is already pushing the limit of what can be cost effectively engineered at this point in time, so why expect 30% OC on a card that is already pushed to the limit when you pull it out of the box.

And someone send some case fans to France, ffs, they seem to have run out.

ignorant might of been the first word to come to mind reading this.
Reviews are exactly that. To tell the good and the bad. IDK WHERE in the world you think that its ok for a company ANY company to produce a product that isn't reliable and get called out for it. Thats just a problem waiting to happen. How could you support a company that could possibly screw you over and who wants to rma a card? really spend another $20 in shipping wait 3 weeks to a year on some company's trying to get a card back
Your reasoning is why you aren't a reviewer nor would be getting a card for free. Nobody wants biased reviews, and yelling about how happy you are for getting a free card sounds like you'd praise it and ignore the faults of the card.
On top of that if you had $700 to spend on the card that you worked hard for..and it dies then ur cardless your telling me thats OK..and not just one out of a million or even a 1000. We are talking 10+ cards that i know of that have died so far..Thats just way too many and these are people running these cards for days..you really think this type of quality can be trusted to last year(s)...i wouldnt chance my hard earned cash on it.

[BababooeyHTJ]

Time will tell, no use crying the sky is falling on day one, is it something to be concerned for, sure, is it a definitive wide spread product issue, too early to really say.

I mean I can't quantify how many burnt vrm's are out there from a handfull of online folks in the hundreds of thousands of units that are actually put into the market. If these issue where so widespread even on other products there wouldn't be any vendors that could afford to sell the product nor could nvidia cover excessive claims.

Is Nvidia being cheap on the vrm, I don't know, are people being too aggressive with what they expect, I don't know.

What I am fairly sure of is if there is an issue now that will potentially lead to excessive warranty claims it will be rectified in short order whether it is bios, driver, or hardware implemented.

570's burn up at even stock.

This is what happens when you start cheaping out on VRM components to avoid powering the capabilities you've created.

I agree, people seem to forget that it's the wattage that kills not necessarily just the voltage. I beat the crap out of my GTX280, that thing overclocked like a beast with nothing but some small enzotech heatsinks on the vrms. Now that card had a quality power regulation. With proper overcurrent and thermal protection like we saw on 4870/90 we shouldn't be seeing all of these pictures of burnt up vrms on 570.

All that I am going to say about 590 is that I don't recall a product launch with more than one report of a card dieing as soon as the NDA is lifted.

It also makes me wonder how TPU could rate this card 70%.

[Postmodum]

hmmm 10+ cards?

I don't know if it's that many ... what i see is people saying "One more died" and posting a link that, in fact, is a foreign website that is talking about the sweclockers video :P eheheh

But i must say that i don't recall any card launch with all this negative buzz around it ...

Only for that, Nvidia ... Fail! We're talking about the absolute flagship product ...

Yes it has potential ... but worthless if you can't control it

[XRL8]

That link on a polish website was linking to sweclockers video because they own sample died too.

[Stukov]

the mainboard is 80-90C hot?
and the scale goes to 90C and then the vga is 110C? 0_o
i think the didnt calibrate the thing properly tbh...

Heat rises...the heat from 120C area is not only heating up the mainboard behind it, but the air above it. It is measuring 80-90C for the air above it.

[Mechanical Man]

the mainboard is 80-90C hot?
and the scale goes to 90C and then the vga is 110C? 0_o
i think the didnt calibrate the thing properly tbh...

Softwares I have used with thermal cam's let u choose colouring scale. It has nothing to do with actual temperature measured in one point. As colour scale can end at 90C as in that pic looks like, but u can use the software to look at numerical values at points. Colour is only for visualization.

In short, Dont mix colour scale and actual temp data measured.

[K404]

nVidias poor in-house validation + testing has probably killed this card.

Who is gonna feel confident spending £570 on a GPU that has seen so many deaths on launch week?

At this level, component confidence means so much, nV might have just lost this round to ATI and there is probably NOTHING they can do about it..... thats bad.

[STEvil]

Softwares I have used with thermal cam's let u choose colouring scale. It has nothing to do with actual temperature measured in one point. As colour scale can end at 90C as in that pic looks like, but u can use the software to look at numerical values at points. Colour is only for visualization.

In short, Dont mix colour scale and actual temp data measured.

Question: Are thermal cam's confused by EMI?

[zanzabar]

Question: Are thermal cam's confused by EMI?

no, maybe if they are in a strong field but they go get a target in an em field.

[tajoh111]

nVidias poor in-house validation + testing has probably killed this card.

Who is gonna feel confident spending £570 on a GPU that has seen so many deaths on launch week?

At this level, component confidence means so much, nV might have just lost this round to ATI and there is probably NOTHING they can do about it..... thats bad.

Thats sort of true. But it will all depend on sales. But considering the limited amount of gtx 590's supposedly, they probably have that round too.